The Astrophysical Journal, 789:114 (14pp), 2014 July 10 doi:10.1088/0004-637X/789/2/114 C 2014. The American Astronomical Society. All rights reserved. Printed in the U.S.A.

THE LICK–CARNEGIE SURVEY: b—A - PLANET ORBITING A NEARBY

Jennifer Burt1, Steven S. Vogt1, R. Paul Butler2, Russell Hanson1, Stefano Meschiari3, Eugenio J. Rivera1, Gregory W. Henry4, and Gregory Laughlin1 1 UCO/, Department of Astronomy and Astrophysics, University of California at Santa Cruz, Santa Cruz, CA 95064, USA 2 Department of Terrestrial Magnetism, Carnegie Institute of Washington, Washington, DC 20015, USA 3 McDonald Observatory, University of Texas at Austin, Austin, TX 78752, USA 4 Center of Excellence in Information Systems, Tennessee State University, Nashville, TN 37209, USA Received 2014 February 24; accepted 2014 May 8; published 2014 June 20

ABSTRACT Precision radial velocities from the Automated Planet Finder (APF) and Keck/HIRES reveal an M sin(i) = 18 ± 2 M⊕ planet orbiting the nearby M3V GJ 687. This planet has an P = 38.14 days and a low . Our Stromgren¨ b and y photometry of the host star suggests a signature with a period of P = 60 days. The star is somewhat chromospherically active, with a spot filling factor estimated to be several percent. The rotationally induced 60 signal, however, is well separated from the period of the variations, instilling confidence in the interpretation of a Keplerian origin for the observed velocity variations. Although GJ 687 b produces relatively little specific interest in connection with its individual properties, a compelling case can be argued that it is worthy of remark as an eminently typical, yet at a distance of 4.52 pc, a very nearby representative of the galactic planetary census. The detection of GJ 687 b indicates that the APF is well suited to the discovery of low-mass planets orbiting low-mass in the as yet relatively un-surveyed region of the sky near the north celestial pole. Key words: planets and : detection – planets and satellites: gaseous planets – stars: individual (GL687) Online-only material: color figures

1. INTRODUCTION definitive statement that is couched in planetary as well as in planetary radii. Figure 1 shows the current distribution The Copernican principle implies that the , and, by of reported planets and planetary candidates orbiting primaries extension, the , do not hold a central or specifically with M < 0.6 M, which we adopt as the functional border favored position. This viewpoint is related to the so-called between “M-type” stars and “K-type stars.” mediocrity principle (Kukla 2010), which notes that an item The census of low-mass planets orbiting low-mass primaries drawn at random is more likely to come from a heavily populated can be accessed using a variety of techniques. For objects category than one which is sparsely populated. near the bottom of the , it appears that These principles, however, have not had particularly apparent photometry from either ground (Charbonneau 2010) or space success when applied in the context of extrasolar planets. Mayor (Triaud et al. 2013) offer the best prospects for planetary et al. (2009) used their high precision Doppler survey data to discovery and characterization. For early to mid M-type dwarfs, deduce that of order 50% (or more) of the chromospherically there is a large enough population of sufficiently bright primaries quiet main-sequence dwarf stars in the solar neighborhood are that precise Doppler detection (see, e.g., Rivera et al. 2010) accompanied by a planet (and in many cases, by multiple can play a lead role. For the past decade, we have had a planets) with M sin(i)  30 M⊕, and orbital periods of P< sample of ∼160 nearby, photometrically quiet M-type stars 100 days. Taken strictly at face value, this result implies that our under precision radial velocity (RV) surveillance with the Keck own solar system, which contains nothing interior to ’s telescope and its HIRES spectrometer. In recent months, this P = 88 day , did not participate in the ’s dominant survey has been supplemented by data from the Automated mode of planet formation. Yet the eight planets of the solar Planet Finder (APF) telescope (Vogt et al. 2014b). Here, we system have provided, and continue to provide, the de facto present 16.6 yr of Doppler velocity measurements for the nearby template for most discussions of planet formation. M3 dwarf GJ 687 (including 122 velocity measurements from Indeed, where extrasolar planets are concerned, M dwarfs Keck, 20 velocity measurements from the APF, and 5 velocity and mediocrity appear to be effectively synonymous. Recent measurements made with the Hobby–Eberly Telescope (HET)) observational results suggest that low-mass planets orbiting low- and we report the detection of the exoplanet that they imply. We mass primaries are by no means rare. Numerous examples of use this discovery of what is a highly archetypal representative planets with Mp < 30 M⊕ and M dwarf primaries have been of a planet in the Milky Way—in terms of its parent star, its reported by the Doppler surveys (e.g., Butler et al. 2004; Mayor planetary mass, and its orbital period—to motivate a larger et al. 2009, and many others), and the Kepler Mission has discussion of the frequency of occurrence, physical properties, indicated that small planets are frequent companions to low- and detectability of low-mass planets orbiting M-type stars. mass stars. For example, Dressing & Charbonneau (2013) report The plan for this paper is as follows. In Section 2, we describe that among dwarf stars with Teff < 4000 K, the occurrence rate of the physical and spectroscopic properties of the red dwarf host = +0.04 star Gliese 687. In Section 3, we describe our RV observations of 0.5 R⊕

1 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al.

Figure 1. Population diagram for currently known extrasolar planets orbiting stars with reported masses Mstar < 0.6 M. Green circles: planets securely Figure 2. H-R diagram with GJ 687’s position indicated as a small open circle. detected by the radial velocity method (either with or without photometric Absolute magnitudes, M, are estimated from V band apparent magnitudes and = transits). Red circles: the regular satellites of the Jovian planets in the solar Hipparcos distances using M V +5log10(d/10 pc). All 956 stars in our system. Gray circles: Kepler candidates and objects of interest. Radii for these catalog of radial velocity measurements (for which more than 20 Doppler candidate planets, as reported in (Batalha et al. 2013), have been converted measurements exist) are shown, color-coded by their B−V values, with point to masses assuming M/M⊕ = (R/R⊕)2.06 (Lissauer et al. 2011), which is areas sized according to the number of observations taken. obtained by fitting the masses and radii of the solar system planets bounded in (A color version of this figure is available in the online journal.) mass by Venus and Saturn, which may be a rather naive transformation given the startling range of observed radii for planets with masses between Earth and . Venus, Earth, and are indicated on the diagram for comparison purposes. Data are from www..org, accessed 2014 December 1. (A color version of this figure is available in the online journal.) observations, along with an analysis that assesses our confidence in the detection. In Section 5, we describe our photometric time series data for the star, which aids in the validation of the planet by ruling out spot-modulated interpretations of the Doppler variations. In Section 6, we discuss the ongoing refinement of the planet– correlation for low-mass primaries, in Section 7, we discuss the overall statistics that have emerged from more than 15 yr of precision Doppler observations of M dwarf stars with the Keck Telescope, and in Section 8 we conclude with an overview that evaluates the important future role of the APF telescope in precision velocimetry of nearby, low-mass stars.

2. GJ 687 STELLAR PARAMETERS Figure 3. Average value of the S-index against the standard deviation of the S-index for all the stars in the Lick–Carnegie database. Stars with M < 0.6 M ◦ Gliese 687 (LHS 450, BD+68 946) lies at a distance, are colored red. GJ 687 is shown as an orange circle in the midst of this d = 4.5 pc, is the 39th nearest known stellar system, and is population, showing that it is a somewhat active star. The areas subtended by the closest star north of +60◦ . Figure 2 indicates the individual points are, in all cases, proportional to the number of Doppler velocity observations that we have collected of the star (with systems above an GJ 687’s position in the color–magnitude diagram (CMD) for upper bound of 250 observations receiving the same point size). stars in the Lick–Carnegie Survey’s database of Keck obser- (A color version of this figure is available in the online journal.) vations. Due to its proximity and its brightness (V = 9.15, Ks = 4.548), Gliese 687 has been heavily studied, and in par- ticular, the CHARA Array has recently been used to obtain 3. RADIAL VELOCITY OBSERVATIONS direct interferometric angular diameter measurements for the star. Boyajian et al. (2012) find R/R = 0.4183 ± 0.0070, and Doppler shifts from both the Keck (122 observations) and derive L/L = 0.02128 ± 0.00023, Teff = 3413 K, and use APF (20 observations) platforms were measured, in each case, the mass–radius relation of Henry & McCarthy (1993) to obtain by placing an iodine absorption cell just ahead of the spectrome- M/M = 0.413 ± 0.041. As illustrated in Figure 3, Gliese ter slit in the converging beam of stellar light from the telescope 687’s mean Mt. Wilson S value (calculated per Wright et al. (Butler et al. 1996). The forest of iodine lines superimposed 2004) and the dispersion of its S-index measurements from our on the stellar spectra generates a wavelength calibration and en- spectra indicate that it has a moderate degree of chromospheric ables measurement of each spectrometer’s point-spread function activity. This conclusion is in concordance with our long-term (PSF). The RVs from Keck were obtained by operating HIRES photometric monitoring program, which also indicates that the at a spectral resolving power R ∼ 70,000 over the wavelength star is somewhat active. range of 3700–8000 Å, though only the region 5000–6200 Å

2 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al.

Figure 4. Lomb–Scargle periodograms for combined radial velocity measure- Figure 5. Phased radial velocity model for planet b, folded at the P = 38.14 day ments of GJ 687 from the HET, Keck, and APF . The horizontal lines orbital period. The blue points correspond to Keck data points, green points are from top to bottom represent false alarm probabilities of 0.01%, 0.1%, and 1.0% APF data, and the red points are HET data. The vertical dashed lines demarcate respectively. the extent of unique data. (A color version of this figure is available in the online journal.) containing a significant density of iodine lines was used in the present Doppler analysis (Vogt et al. 1994). The APF measure- ments were obtained over a similar spectral range, but at a higher spectral resolving power, R ∼ 108,000. For each spectrum that was obtained, the region containing the iodine lines was divided into ∼700 chunks, each of ∼2 Å width. Each chunk produces an independent measure of the wavelength, PSF, and Doppler shift. The final measured velocity is the weighted mean of the velocities of the individual chunks. All RVs have been corrected to the solar system barycenter, but are not tied to any absolute velocity system. As such, they are “relative” velocities, with a zero point that can float as a free parameter within an overall system model. The internal uncertainties quoted for all the RV measurements in this paper reflect only one term in the overall error budget, and result from a host of errors that stem from the characterization and determination of the PSF, detector imperfections, optical Figure 6. Lomb–Scargle periodogram of the radial velocity residuals to the fit aberrations, consequences of undersampling the iodine lines, given in Table 2 plotted in black, and the Lomb–Scargle periodogram of the and other effects. Two additional major sources of error are Mt. Wilson S-index values plotted behind in red. photon statistics and stellar “jitter.” The latter varies widely from (A color version of this figure is available in the online journal.) star to star, and can be mitigated to some degree by selecting magnetically inactive older stars and by time-averaging over the star’s unresolved low-degree surface p-modes. All observations Using Levenberg–Marquardt optimization, we obtained a best-fit Keplerian model for the system. This fit, which assumes in this paper have been binned on 2 hr timescales. In addition ◦ ◦ to the RVs that we have obtained at Keck and APF, we also use i = 90 and Ω = 0 for the planet, is listed in Table 2. five Doppler measurements obtained by Endl et al. (2003)atthe The phased RV curve for the planet in Table 2 is shown in HET located at McDonald Observatory. These RV observations Figure 5. A power spectrum of the residuals to our one-planet fit are presented in the Appendix. is shown in Figure 6 and indicates no significant periodicities. Also shown in this figure is a periodogram of our Mt. Wilson S-index measurements from the spectra, which are a proxy for 4. THE BEST FIT SOLUTION the degree of spot activity on the star at a given moment. None The combined RV data sets show a root-mean-square (rms) of the peaks in the periodogram of S-index values coincide with scatter of 7.58 m s−1 about the mean velocity. This scatter is the peak that we suspect to be a planet. 2 = measured after we have applied best-fit telescope offsets of The reduced chi-squared statistic for our fit is χred 18.55 0.64 m s−1 for Keck, −1.71 m s−1 for APF, and 1.27 m s−1 and results in a fit with a combined rms of 6.16 m s−1 and −1 for HET. estimated excess variance of σjitter = 5.93 m s (the estimate of A Lomb–Scargle periodogram of the 149 velocity measure- the jitter that is required to bring the reduced chi-squared statistic ments of GJ 687 is shown in Figure 4. False alarm probabil- of the fit down to unity). This value accounts for variance in ities (FAPs) are calculated with the bootstrap method, as de- both the stellar signal and from the telescope itself, though for scribed in Efron (1979), iterating 100,000 times for a minimum a moderately active star such as GJ 687 a stellar jitter of order −1 probability of Pfalse < 1e − 5 as easily met by the tallest Pb = 6.0 m s is reasonable, and could account for the majority of 38.14 day peak in Figure 4. This signal in the data is modeled the observed variance. as a Mb sin(i) = 0.06 MJ planet with an orbital eccentricity, In order to compute parameter uncertainties for our orbital fit, eb = 0.04. we implement a Markov Chain Monte Carlo algorithm (Ford

3 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al.

Figure 7. Smooth scatter plots of parameter error correlations for our Markov Figure 8. Orbit of the proposed planetary companion to GJ 687. The larger chain. In each case, the best-fit model is indicated with a small red dot, and red point corresponds to the location of the planet at the initial observation the density of models within the converged portion of the chain is shown as a , HJD 2450603.97. The line from the origin corresponds to the planet’s blue-toned probability distribution function. The diagonal line of entries shows periastron. For the geometry plotted, transits, should they occur, would happen the marginalized distribution for each parameter of the one-planet model. when the planet traverses the positive y-axis. The light lines are 100 of (A color version of this figure is available in the online journal.) the planet drawn from the converged segment of the Markov Chain. The red dot in the center of the diagram corresponds to the size of the star when drawn to scale. The small black dot next to the distance scale bar indicates the size of the planet when drawn to scale, and assuming it has RP = RNep. 2005; Ford & Gaudi 2006; Balan & Lahav 2009; Meschiari (A color version of this figure is available in the online journal.) et al. 2009; Gregory 2011). The MCMC algorithm returns a chain of state vectors, ki (a set of coupled , e.g., period, mass, etc. and the three velocity offset parameters). The goal of the Markov Chain calculation is to generate an equilibrium distribution proportional to exp[χ 2(k)]. We adopt non-informative priors on all parameters (and uniform in the log for masses and periods). The resulting error correlations are shown in Figure 7, and a set of 100 states drawn randomly from the converged chain are shown in the orbital diagram in Figure 8. The error correlation diagram indicates that all parameters are well determined, save the usual degeneracies between mean anomaly and ω for the low-eccentricity orbit. The distribution of the residuals relative to the best-fit model shows no evident pathologies. Indeed, a quantile–quantile plot (shown in Figure 9) indicates that the distribution of residuals is well described by a normal distribution. We note that the smaller scatter of points obtained with the APF telescope could be a consequence of the fact that they were all taken within a Δt = 140d period, and thus sample only one segment of the stellar activity cycle. A potentially significant challenge to correctly identifying the orbital period of a proposed exoplanet arises from the discrete and uneven sampling inherent in RV surveys. The spacing of observations leads to increased noise and the presence of aliases within the star’s periodogram which can be mistaken for a true orbital signature. For a real signal occurring at a frequency fplanet we expect alias signatures at f = fplanet ± nf sampling where n is an integer. In order to aid confirmation that the periodic signal Figure 9. Quantile–quantile plot for the velocity residuals to the one-planet model fit. Adherence of the points to the lines indicate the degree to which the we observe is actually a planetary signature, we must be able radial velocities from the two telescopes conform to a normal distribution. APF to calculate where aliases due to our observing cadence will points are shown in green, Keck points are shown in blue, and HET points are occur, and then verify that they are not the source of the signal. shown in red. The aliases are determined using a spectral window function as (A color version of this figure is available in the online journal.)

4 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al. Arizona. The T12 APT is one of several TSU automatic tele- scopes operated at Fairborn (Henry 1999; Eaton et al. 2003). It is equipped with a two-channel precision photometer that em- ploys a dichroic filter and standard Stromgren¨ b and y filters to separate the two passbands and two EMI 9124QB bi-alkali photomultiplier tubes to measure the b and y count rates simul- taneously. We observed GJ 687, designated our program star (P), differentially with respect to three neighboring comparison stars: C1 (HD 156295, V = 5.54, B − V = 0.22, F0 IV), C2 (HD 160198, V = 7.65, B − V = 0.46, F2 V), and C3 (HD 161538, V = 7.01, B −V = 0.44, F2 V). A detailed description of the observing sequence and the data reduction and calibration procedures are given in Henry (1999). We computed all pairwise differential magnitudes P − C1, P − C2, P − C3, C3 − C2, C3 − C1, and C2 − C1 in both the b and y passbands, corrected them for atmospheric extinc- tion, and transformed them to the standard Stromgren¨ photo- Figure 10. Window function calculated from all radial velocity observations of metric system. Observations with internal standard deviations GL 687. While several peaks exist due to aliasing effects from our data’s time greater than 0.01 mag were discarded to remove data taken stamps, none of them coincide with the locations necessary to create a peak in in non-photometric conditions. Intercomparison of the six sets the periodogram at our best fit period of P = 38.14 days. of differential magnitudes demonstrated that HD 156295 (C1) is a low-amplitude variable while both HD 160198 (C2) and defined by Roberts et al. (1987): HD 161538 (C3) are constant to the expected measurement pre- cision. To improve our precision, we combined the separate N differential b and y observations into a single (b + y)/2 “pass- 1 − W(ν) = exp 2πiνtr , (1) band.” We also computed the differential magnitudes of GJ 687 N with respect to the mean brightness of the two good comparison r=0 stars: P − (C2+C3)/2. The standard deviation of the C3 − C2 where N is the total number of observations and t is the date on comparison star differential magnitudes is 0.0020 mag, which which they were taken. Plotting this function will result in peaks we take to be the precision of a single measurement. that are due solely to the sampling cadence of the data. Because A total of 606 nightly measurements in the five observing our observations are constrained by when the star is visible in seasons survived the cloud-filtering process. These data are the night sky, and because Keck Telescope time is allocated plotted as P − (C2+C3)/2 differential magnitudes in the top to Doppler surveys primarily when the Moon is up, we expect panel of Figure 11. The five individual observing seasons are aliases at periods of 1 solar day, 1 sidereal day, 1 synodic month plotted in the remaining panels. The standard deviations for and 1 sidereal . Examining the window function in Figure 10 the yearly light curves are given in each panel. These range we do see peaks resulting at these periods, but careful analysis from 0.0049 to 0.0092 mag, compared to the measurement of the periodogram for our RV observations shows no evidence precision of 0.0020 mag. Gaps of 10–12 weeks in the yearly of strong signals occurring at the locations necessary for our light curves for 2009 through 2012 are due to southern Arizona’s P = 38.14 day signal to be a potential alias instead of a true July–September rainy season when good photometry is not Keplerian signature. possible. With an apparent Ks-band magnitude of 4.54, Gliese 687 is Low-amplitude variability is seen in GJ 687 during each brighter (in the near ) than all known hosts of transit- observing season, resembling light curves typical of modestly ing extrasolar planets other than 55 Cancri. As a consequence, active stars with spot filling factors of a few percent (see, transits by Gliese 687’s planetary companion (which has an e.g., Henry et al. 1995). The 2010 light curve has the largest ∼ equilibrium , Teq ∼ 260 K), were they to occur, amplitude variability ( 0.03 mag) and reveals cyclic variation would be of substantial scientific value. In particular, transmis- with a time scale of ∼60 days. The other light curves have sion spectroscopy with James Webb Space Telescope (JWST) lower amplitudes and include cyclic variations of ∼60 and also would give insights into what is likely a dynamic and chemi- ∼30 days. These year-to-year and cycle-to-cycle variations are cally rich planetary . The a priori geometric transit also typical of modestly active stars. We interpret the 60 day probability for Gliese 687 b, however, is a scant Ptr = 1.2%, and variability as the signature of the star’s rotation period and as we describe below, there is no evidence that transits occur. the 30 day variability as a sign of spot activity on opposite With M sin(i) = 19 M⊕, the currently observed mass–radius hemispheres of the star. This slow rotation rate is in agreement range for exoplanets indicates that the planetary radius, R could with the work done by Jenkins et al. (2009) which reports an p −1 credibly range from Rp ∼ 0.2RJup to Rp ∼ 0.6 RJup, implying upper rotational velocity of 2.8kms for GL 687. potential transit depths in the d = 0.2% to d = 2% range. Frequency spectra of the complete 2009–2013 data set and of the 2010 data alone are shown in the top and bottom panels 5. PHOTOMETRIC OBSERVATIONS of Figure 12 respectively. The rotational modulation signal is seen most clearly in the 2010 data, which matches up with the During the 2009–2013 observing seasons, we acquired a to- most coherent light curve in Figure 11. Therefore, we take the tal of 866 photometric observations of GJ 687 on 519 nights 58.48 ± 1.0 days signal as our best measurement of the star’s with the Tennessee State University (TSU) T12 0.80 m auto- rotation period. Inspection of the 2010 photometric segment matic photoelectric telescope (APT) at Fairborn Observatory in of Figure 11 clearly shows the overall 60 day modulation

5 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al.

Figure 12. Lomb–Scargle periodogram of the photometric observations of GJ 687. In the combined data set, the maximum observed power occurs at 58.48 days (top panel). However when we consider only data obtained in 2010, where the rotational modulation is most clearly exhibited, we find that maximum power occurs at P = 61.73 days (bottom panel). We identify this periodicity with the rotational period of the star.

each yearly light curve using the method described in Henry et al. (2001). We removed three to six frequencies from each light curve until each set of residuals approached the precision of a single observation. The residuals from all five observing seasons are plotted in the top panel of Figure 13, phased with the 38.14 day best-fit planetary orbital period and a time of mid transit computed from the orbital parameters. The vertical bar represents the 0.0022 mag standard deviation of the residuals from their mean, very close to the measurement precision given Figure 11. Photometric data taken of GJ 687 over 5 yr. The top panel shows the above. A sine fit to the phased data gives a formal semi- total data set with information regarding observations and standard deviation amplitude of just 0.00011 ± 0.00012 mag. Since none of the for each year. The bottom panel gives a closer look at the data separated by year. frequencies removed from the yearly light curves were similar (A color version of this figure is available in the online journal.) to the orbital frequency or its harmonics, this result limits any periodic brightness variability of the star on the observed that generates the periodogram peak. Departures from perfect RV period to a very small fraction of one millimagnitude periodicity are presumably caused by the evolution of the spot (mmag). This rules out the possibility that the 38.14 day RV activity on the surface of the star. variations in GJ 687 are induced by stellar activity, as has been Finally, we search for transits of GJ 687 b by first removing documented in somewhat more active stars, for instance, by the spot variability from each of the yearly light curves. We Queloz et al. (2001), Paulson et al. (2004), and Boisse et al. do this by successively subtracting multiple frequencies from (2012). Instead, this lack of photometric variability confirms

6 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al.

Figure 14. For each of the 10,402 potential systems in our Markov chain, we check the predicted transit times against our photometric observations. If a Figure 13. Top panel: filtered differential photometric measurements for Gliese photometric data point lies within the transit window of a particular member of 687 folded at the best-fit planetary period, P = 38.14 days. A light curve the Markov chain, we assign a value to that point which is cosine-weighted by model for a centrally transiting Neptune-sized planet is shown. The vertical its distance from the predicted time of central transit. The sum of these values is error bar indicates the 0.002 mag photometric precision. The horizontal error mapped onto the color of the points in the diagram. The phase of the points, as bar shows the 1σ uncertainty on the time of a central transit. Bottom panel well as the vertical gray bar spanning the predicted 3 hr central transit duration shows a magnified view of the folded photometric data in the vicinity of the are for our best fit model given in Table 2. predicted time of central transit. (A color version of this figure is available in the online journal.)

Table 1 Table 2 Stellar Parameters for Gliese 687 One-planet Model for the GJ 687 System Parameter Value Reference Parameter Best fit Errors Spectral type M3 V Rojas-Ayala et al. (2012) Period (days) 38.14 (0.015) Mass (M) 0.413 ± 0.041 Boyajian et al. (2012) Mass (MJ) 0.058 (0.007) Radius (R) 0.4183 ± 0.0070 Boyajian et al. (2012) Mass (M⊕) 18.394 (2.167) (L) 0.0213 ± 0.00023 Boyajian et al. (2012) Mean anomaly (deg) 234.62 (87.962) Distance (pc) 4.5 ± 0.115 Rojas-Ayala et al. (2012) Eccentricity 0.04 (0.076) B−V 1.5 Simbad Longitude of periastron (deg) 359.43 (120.543) V mag. 9.15 Rojas-Ayala et al. (2012) Semi-major axis (AU) 0.16353 (0.000043) J mag. 5.335 Cutri et al. (2003) Time of periastron (JD) 2450579.11 (9.32) H mag. 4.77 Cutri et al. (2003) RV half amplitude (m s−1) 6.43 (0.769) K mag. 4.548 Cutri et al. (2003) First observation epoch (JD) 2450603.97 Avg. S-index 0.811 This work Velocity offsets σ 0.096 This work S-index Keck/HIRES 0.64 m s−1 (0.63) P (days) 61.8 ± 1.0 This work rot APF/Levy −1.71 m s−1 (1.68) T (K) 3413 ± 28 Boyajian et al. (2012) eff HET 1.27 m s−1 (0.98) χ 2 18.55 rms that the RV variations in GJ 687 result from true planetary Keck/HIRES 6.62 m s−1 reflex motion. APF/Levy 3.95 m s−1 The photometric observations within ±0.13P of mid-transit HET 2.44 m s−1 are replotted with an expanded scale in the bottom panel of Jitter 5.93 m s−1 Figure 13. The solid curve shows the predicted phase, depth (assuming Neptune-like density), and duration of a central Notes. All elements are defined at epoch JD = 2450603.97. Uncertainties transit, computed from the stellar radius in Table 1 and the are reported in parentheses. orbital elements in Table 2. The horizontal error bar under the ± predicted transit time gives the 1σ uncertainty in the timing in the orbit, the potential transit duration, the potential size of the of the transit. The photometric observations when filtered using planet, and the error in the photometric filtering, we recommend the Henry et al. (2001) procedure described above, and when that continued photometric monitoring be carried out to confirm = folded at the P 38.14 day best-fit period for the planet, give that transits do not occur. no indication that transits occur. We note, however, that the Markov Chain models generate a five-day window for possible 6. METALLICITY transits, and so a more conservative approach is also warranted. In Figure 14, we plot the unfiltered photometric data, indicating Gliese 687 appears to have a slightly sub-solar metallicity. the range of photometric points that potentially could have been Rojas-Ayala et al. (2012)useNai,Cai, and H2O-K2 calibrations affected by transits were they to occur. Because of uncertainties to estimate [Fe/H] =−0.09 for Gliese 687, whereas the

7 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al. M dwarf metallicity calibration of Schlaufman & Laughlin 2010 yields a value [Fe/H] =−0.02. The connection between the detectable presence of a giant extrasolar planet and the metallicity of the host star was noticed soon after the first extrasolar planets were detected (Gonzalez 1997), and has been studied in many previous works, see, e.g., Fischer & Valenti (2005); Sousa et al. (2011). For M dwarfs, recent work, such as that by Neves et al. (2013), suggests that the stellar metallicity correlation holds robustly for M dwarf primaries, but that for planets with mass, Mp  20 M⊕, no correlation is found with host star metallicity, and indeed, Neves et al. (2013) and Jenkins et al. (2013) report a hint of anti-correlation between the presence of a low-mass planet and host star [Fe/H]. Our detection of a Neptune-mass companion to Gliese 687, and our Lick–Carnegie database of Doppler measurements of M dwarf stars provides an opportunity to revisit this topic. Our database of RV observations taken at the Keck Telescope contains 142 M-type stars with the necessary spectral informa- tion to assess metallicity, 17 of which are known to host planets published in the peer-reviewed literature. We break the planet- hosting stars into two subgroups based on their masses—s- tars with M sin(i) planets less than 30 MEarth are described as Neptune hosting while stars with M sin(i) planets greater than 30 MEarth are listed as Jupiter hosting. We replicate the procedure of Schlaufman & Laughlin (2010) and examine how horizontal distance from a field M dwarf main sequence in a MKs versus (V − Ks ) CMD correlates with metallicity, as noted, e.g., by Baraffe et al. (1998). The top panel of Figure 15 displays all of the Lick–Carnegie survey M dwarf stars plotted in MKs versus (V − Ks ), with gray dots denoting survey stars without known planets, red dots denoting survey stars that host “Neptune-mass” planets and blue dots representing the survey stars that host “Jupiter-mass” planets. It can be seen that most planet hosting stars fall to the right of the field M dwarf main sequence pre- sented in Johnson & Apps (2009) (black line), which is taken Figure 15. Top panel: location of the 142 M dwarfs from the Lick–Carnegie =− radial velocity survey. Stars known to host Jupiter-mass planets are plotted to be a [Fe/H] 0.17 isometallicity contour in this CMD. In in blue, those known to host twice-Neptune M sin(i) (or smaller) planets are order to quantify the likelihood that a star’s horizontal distance plotted in red, and non-planet hosting survey M dwarfs are plotted in gray. The from the isometallicity contour is related to its propensity to host field M dwarf main sequence from JA09 is shown as a black line and the arrows planets, we compare the distances for our actual planet-hosting affixed to each point represent that survey star’s . Bottom panel: distributions generated via Monte Carlo simulations of the cumulative sample stars with randomly drawn samples from the collection of M distance of field M dwarfs (Σ as defined in Equation (2)) from the M dwarf main dwarfs in the survey. sequence used by Johnson & Apps (2009). The points plotted on top of each We characterize the position of each M dwarf by obtaining curve in the bottom panel represent the actual cumulative distance from the MS for our planet hosting and field star samples. V-band and Ks photometry and then using them to calculate the distance statistic Σ: (A color version of this figure is available in the online journal.)

n Σ = (V − K ) − (V − K ) . (2) given no correlation between whether the star hosts an exoplanet s i s iso − − i=1 and its location in the (V Ks ) (MKs )CMD. Our results show that for hosts of Jupiter mass planets, Σ = = +0.13 To determine if the Σ of our known planet hosting subgroups 2.359, which corresponds to a probability of p 0.053−0.04 is significant or, alternatively, if it could be produced by chance, that the stars’ cumulative distance from the isometallicity we make use of a Monte Carlo simulation that calculates contour occurred by chance. For the Neptune hosts, we find that Σ = = +0.24 the cumulative sample distance of survey M dwarfs from the 1.113 leading to p 0.452−0.23, and for the combination Σ = = +0.09 field M dwarf main sequence presented by Johnson & Apps of all planet hosts, we obtain 3.473 or p 0.0775−0.05.The (2009). For the simulation, we randomly select a subset of distributions resulting from the Monte Carlo simulation and the M dwarfs from the Lick–Carnegie field star list, setting the locations of the actual planet hosting stars Σ values are displayed sample size equal to the number of M dwarfs known to host in the bottom panel of Figure 15. The points plotted on top of either Jovian or Neptune mass planets. Then we compute the each curve in Figure 15 represent the actual cumulative distance cumulative horizontal distance of those stars from the field of our planet hosting star samples from the field M dwarf MS. M dwarf MS, where stars to the right of the MS add their Our results thus indicate that the planet–metallicity correlation distance to the sum and stars to the left of the MS subtract is robust for M dwarf hosts of planets with M>30 M⊕, but that their distance. We repeat this process 10,000 times to determine at smaller masses there is, at present, no evidence a correlation the distribution of cumulative horizontal distances from the MS exists.

8 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al. 7. PLANET RECOVERY Table 3 Planets Found from 179 Local M Dwarf Stars The Lick–Carnegie exoplanet survey and its predecessors Star Planet Source Per MJup Ecc have carried out a long-term monitoring program of the brightest (days) M dwarf stars in the sky. Our database of observations contains 159 stars that have more than 10 observations apiece, and which, HD 285968 GJ 176 b Calc 8.775 0.031 0.0 additionally, have median internal uncertainty σ<10 m s−1. GJ 176 b Pub 8.7832 0.026 0.0 Within this group, there is a subset with extensive data sets. HIP 22627 GJ 179 b Calc 2249.994 0.829 0.197 For example, 11 stars have N>100 observations and median GJ 179 b Pub 2288 0.824 0.210 − internal uncertainties σ<13 m s 1. A question of substantial GJ 1214 GJ 1214 b Calc 2.732 0.024 0.11 interest, therefore, is the degree to which the observations taken GJ 1214 b Pub 1.580 0.0204 0.0 to date have probed the true aggregate of planetary companions GL 317 GL 317 b Calc 692.965 1.164 0.0 to the M dwarf stars in our survey. GL 317 b Pub 692.9 1.18 0.193 The effort required to obtain the existing data has been sub- GL 382 GL 382 Calc 36.380 0.071 0.0 stantial. Among the M-type stars alone, our database contains a total of 5,468 velocity measurements from Keck I, totaling GL 388 GL 388 Calc 2.226 0.143 0.06 2,579,862 s (29.86 days) of on-sky integration. Overheads, in- GL 433 GL 433 b Calc 7.3699 0.016 0.2 cluding the acquisition of high S/N spectra, CCD readout time, GL 433 b Pub 7.371 0.0182 0.080 and weather losses, add materially to this time investment. Fur- GL 625 GL 625 Calc 21.316 0.032 0.33 thermore, the distribution of total observing time allotted to the GL 876 GJ 876 b Calc 61.033 1.924 0.0 stars on the list has been highly uneven. Targets such as Gliese GJ 876 c Calc 30.228 0.636 0.01 436 and Gliese 876, which harbor planetary systems of particu- GJ 876 d Calc 15.042 0.117 0.05 lar interest, have received much more attention than the typical GL 876 GJ 876 b Pub 61.117 1.95 0.032 red dwarf in the survey. For example, Gliese 436 has 148 obser- GJ 876 c Pub 30.088 0.612 0.26 vations and Gliese 876 has 204 observations obtained with the GJ 876 d Pub 1.938 0.018 0.207 Keck Telescope. The stars themselves also exhibit a range of GJ 876 e Pub 124.26 0.0392 0.055 chromospheric activity levels. The resulting star-to-star disper- HIP 109388 GJ 849 b Calc 2001.096 1.049 0.0 sion in “stellar jitter” (tantamount to a measurement uncertainty, GJ 849 b Pub 1880 0.83 0.040 σjit) complicates the evaluation of threshold levels for M sin(i) HIP 57050 HIP 57050 b Calc 41.373 0.267 0.291 as a function of orbital period to which planetary companions HIP 57050 b Pub 41.397 0.298 0.314 can be excluded. There are a variety of approaches to the measurement of FAPs HIP 57087 GJ 436 b Calc 2.644 0.067 0.149 GJ 436 b Pub 2.644 0.0737 0.15 in the context of spectral analysis of unevenly sampled data; see, e.g., Baluev (2012) for a recent discussion. A very simple HIP 70890 HIP 70890 Calc 364.742 0.021 0.1 approach is described by Press et al. (1992). For a Gaussian HIP 74995 GJ 581 b Calc 5.368 0.048 0.0 random variable,5 the probability distribution for obtaining a GJ 581 c Calc 12.918 0.017 0.156 peak at frequency ω of Lomb-normalized power (Scargle 1982), GJ 581 d Calc 68.689 0.017 0.4 HIP 74995 GJ 581 b Pub 5.369 0.0499 0.031 PN(ω), is exponential with unit mean. If a data set drawn from measurements of a white noise (Gaussian) distribution supports GJ 581 c Pub 12.918 0.0168 0.070 measurement of M independent frequencies, the probability GJ 581 d Pub 66.640 0.0191 0.250 that no peak exceeds power z (the FAP) is P(PN >z) = HIP 83043 GJ 649 b Calc 592.764 0.267 0.33 1 − (1 − exp−z)M . GJ 649 b Pub 598.3 0.325 0.3 − We adopt a FAP of 10 4, calculated with the above method HIP 85523 GJ 674 b Calc 4.691 0.034 0.19 (and using Monte-Carlo simulations to determine M)asthe GJ 674 b Pub 4.694 0.035 0.2 generic threshold for attributing a given planetary signal to HIP 109388 GJ 849 b Calc 2001.099 1.049 0.14 a given data set. With this detectability threshold, we use GJ 849 b Pub 1880 0.83 0.040 the Systemic Console 2.0 software package (Meschiari et al. HD 204961 GJ 832 b Calc 3440.923 0.639 0.12 2012) to determine the number of readily detectable plan- GJ 832 b Pub 3420 0.644 0.12 ets in our M dwarf data set. A “readily detectable” planet generates a signal that can be isolated algorithmically (and automatically) by straightforward periodogram analysis and Levenberg–Marquardt minimization. The results of this exer- finds no planets. Regarding GJ 667C, 40 observations have been cise are shown in Table 3, which locates signals corresponding made at Keck, and these were used in a characterization of the to 19 previously published planets orbiting 14 separate M dwarf GJ 667C system (Anglada-Escudeetal.´ 2012), however the primaries. The table lists both the published planetary parame- peak planetary signal for this set fell below our FAP threshold ters from exoplanets.org (Wright et al. 2011) and the parameters when utilizing only the Keck data. The bright planet-hosting red calculated by our search. Other than Gliese 667C, there are no dwarfs Gliese 832, 3634, and 3470 all have that are stars on our 159 star list for which a planet has been pub- too far south to be observed from Mauna Kea, and HIP 79431 h m s − ◦   lished by another group, and for which the automated algorithm (R.A. 16 12 41.77, decl. 18 52 31.8) is not on the list of M dwarfs being monitored at Keck. The Kepler mission’s photometric data have been used to 5 Clearly, the generating function for typical radial velocity data sets has infer that small planets orbiting M dwarfs are very common. For non-Gaussian (and unknown) error. False alarm probabilities must therefore be treated with great caution when evaluating the existence of a planet with example, Dressing & Charbonneau (2013) find an occurrence K  σunc.. rate of 0.9 planets per star in the range 0.5 R⊕

9 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al.

Figure 16. Example plots of our synthetic planet recovery around four M dwarf stars. The points represent planets our algorithm found, colored by the false alarm probability for the initial detection. The black lines from bottom to top show radial velocity half amplitudes (K) of 1, 10, and 100 m s−1 . The green line is our minimum detectable K value. (A color version of this figure is available in the online journal.) with P<50 days. Given the existence of this large number detectablity thresholds lie roughly along lines of constant K. of small-radius planets, it is of interest to make a quantitative To determine the smallest K value we could reliably detect for analysis of how deep into the expected population of super- each star, we find the smallest value of K for which a planet Earth-type planets suggested by the Kepler mission the Keck was found for at least 50% of the chosen periods. This median Radial Velocity Survey has probed. To answer this question, we value generates the green lines seen in the figure. The top panel have created synthetic RV data sets that contain test planets, and of Figure 17 shows these minimum K values for each of the which conform with the timestamps, the internal measurement 104 stars that we tested. For clarity, the stars in this figure have uncertainties, and the stellar properties (namely mass) for all been ordered by increasing minimum K, and are colored by the 104 M dwarfs under surveillance at Keck with at least 20 RV number of observations we have for each. observations. To address the error source arising from stellar If a test planet lies in its star’s habitable zone, (defined as jitter, σjit, we use the median value provided in Wright (2005)of the semi-major axis at which the flux received by the planet is −1 3.9 m s as the expected level of σjit for our M dwarf stars. This the solar constant received at Earth) we can ask how large the value is then added in quadrature with the internal uncertainty planet needs to be to be detectable by our RV survey. Figure 17 and applied to the synthetic data set to create a more accurate shows these threshold masses for each of the 104 M dwarf stars representation of the system. which we analyzed. These stars maintain the ordering from the For each of the 104 stars in the Keck survey, we have created top panel, but now have been colored by the mass of the parent 400 synthetic data sets. Each set contains a single planet. The star. We see that while the Keck survey has probed substantially planets are evenly spaced in log period from 2 to 100 days, and into the regime occupied by Neptune-mass planets, it has not evenly spaced in log mass from 1 M⊕ to 1 MJup. We assign a made significant inroads into the super-Earth regime for periods circular orbit to these test planets and assume i = 90◦ and Ω = that are of astrobiological interest. 0◦ in each case. We then calculate the RV each of these planets would induce on a parent star. A Gaussian distribution with 2 2 2 8. DISCUSSION σ = σinternal + σjitter is used to perturb the predicted RV value. Each of the 104×400 synthetic systems is passed to the planet GJ 687 is the second to be detected using search algorithm. Figure 16 shows examples of the returned data from the APF telescope, with the first being HD 141399 planets for 4 of the 104 stars in this experiment including the star b,c,d, and e (Vogt et al. 2014a). APF has successfully navigated of main interest here, GJ 687. In Figure 16 the black lines from its commissioning stage, and, since Q2 2013, it has routinely 1/3 −2/3 bottom to top represent constant K = (2πG/P) MPM acquired science-quality data that presents sub-m/s precision on − values of 1, 10, and 100 m s 1 respectively. As expected, the known RV standard stars (Vogt et al. 2014b). In recent months,

10 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al. key contributions to exoplanet detection and characterization in the coming . Gliese 687 b’s RV half-amplitude, K = 6.4 ± 0.5ms−1,is substantially greater than the current state-of-the-art detection threshold for low-mass planets. The lowest measured value for K in the catalog of Doppler-detected extrasolar planets6 stands at K = 0.51 m s−1 (Dumusque et al. 2012). On the other hand, Gliese 687’s status as one of the nearest stars to the imbues it with a great deal of intrinsic interest. In our view, the relatively recent date for Gliese 687 b’s detection can be attributed both to the substantial amount of stellar-generated RV noise (as evidenced by Figures 3 and 5), but also to its location in , high in the Northern Sky, where APF, along with HARPS North, are the only facilities that can routinely observe at sub-1 m s−1 precision. (As evidenced by the data in this paper, Keck can observe at these high declinations, but at significantly higher expense in comparison to stars lying closer to the celestial equator.) Indeed, Gliese 687’s stellar coordinates (R.A. 17h36m, decl. +68◦) place it very close to the north ecliptic pole, located at R.A. = 18h, decl. = +66◦. This location flags it as a star of potentially great importance for the forthcoming NASA TESS Mission. As currently envisioned (and as currently funded), TESS is a two-year, all-sky photometric survey to be carried out by a spacecraft in a 27 day P/2 lunary resonant orbit. TESS will photometrically monitor ∼500,000 bright stars with a <60 ppm 1 hr systematic error floor. (For reference, a central transit of the Sun by the Earth produces an 86 ppm transit depth.) The northern ecliptic hemisphere will be mapped during the first year of the mission via a sequence of 13 sectors with 27 days of continuous observation per sector. These sectors overlap at the north ecliptic pole, and create an area of ∼1000 deg2 (1/50th of the sky) for which photometric baselines will approach 365 days. Gliese 687 lies at the center of this TESS “overlap zone” (which also coincides with JWST’s continuous viewing zone). Because much longer time series are produced Figure 17. Top panel: the minimum detectable K value for each star in our M in the overlap zone, the highest-value transiting planets found dwarf collection. Bottom panel: assuming a minimum detectable K for each by the mission will emerge from this part of the sky (along with M dwarf star, if a planet was orbiting in that star’s habitable zone, this is the minimum mass that planet could have and still be recovered by our method. the sister segment covering the south ecliptic pole). (A color version of this figure is available in the online journal.) As mentioned above, however, the TESS overlap zone has received relatively little attention from the highest-precision Doppler surveys. About 10,000 target stars from the TESS Dwarf Star Catalog, all with V<12, are present in the overlap zone. This, of course, is far too many stars to survey with the degree of automation for APF has increased substantially. Doppler RV, but there appears to be substantial value inherent in The facility currently works autonomously through an entire monitoring the brightest, nearest, and quietest members of the night’s operations, calibration, and observing program. The APF cohort of TESS overlap stars. The latest estimates (Mayor et al. and its accompanying high-resolution Levy spectrograph to- 2009, 2011; Batalha et al. 2013; Petigura et al. 2013) suggest gether form a dedicated, cost-effective, ground-based precision that ∼50% of main sequence stars in the solar vicinity harbor RV facility that is capable of detecting terrestrial-mass planets M>M⊕ planets with P<100 days. Assuming a uniform at distances from their parent stars at which surface liquid water distribution in period between 5 and 100 days, the average transit could potentially be present. probability for these planets is P ∼ 2.5%, suggesting that of Unlike other highly successful RV facilities, the APF uses order N ∼ 0.5 × 0.025 × 10,000 ∼ 125 low-mass transiting neither image scrambling nor image slicing. With a peak effi- planets (and systems of transiting planets) will be detected by ciency of 15% and typical spectral resolutions of R ∼ 110,000, TESS within the overlap zone. Of these, a small handful, of order the APF represents a critical new resource in the global quest to 5 systems total (and perhaps, with probability P = 1.2%, detect extrasolar planets. Initial speed comparisons indicate that transit including Gliese 687 b) will garner by far the most attention in order to match the signal-to-noise acquired using the Keck from follow-up platforms such as JWST, due to their having telescope/HIRES Spectrograph combination, the APF needs optimally bright parent stars. only a factor of 6 increase in observing time. Since the amor- Our detection of Gliese 687 b suggests that by starting now, tized cost of a night on Keck is ∼77 times more expensive than with a systematic program of Doppler observations of a target a night on the APF, and because 80% of the APF’s nights are reserved for exoplanetary work, the APF (with its sub-m/s preci- sion and dedicated nightly cadence abilities) will likely provide 6 www.exoplanets.org

11 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al. list of ∼200 carefully vetted G, K and M dwarf stars with Table 4 V ∼ 7.5 to V ∼ 10.5 in the 1000 deg2 TESS overlap zone, APF (Continued) can ensure that a precise multi-year Doppler velocity time series JD RV Uncertainty will exist for the most important TESS planet host stars at the (m s−1)(ms−1) moment their transiting planets are discovered. 2452007.029 −3.280 1.970 2452009.077 −7.440 2.020 This research has made use of the Exoplanet Orbit Database 2452061.913 7.450 2.040 and the Exoplanet Data Explorer at exoplanets.org. G.L. ac- 2452062.935 4.490 1.940 2452094.853 2.450 1.550 knowledges support from the NASA Astrobiology Institute 2452096.894 4.560 1.850 through a cooperative agreement between NASA Ames Re- 2452097.982 5.180 1.610 search Center and the University of California at Santa Cruz. 2452127.919 1.940 2.200 S.S.V. gratefully acknowledges support from NSF grants AST- 2452133.737 2.830 2.010 0307493 and AST-0908870. R.P.B. gratefully acknowledges 2452160.872 −6.120 2.170 support from NASA OSS grant NNX07AR40G, the NASA 2452161.821 −6.840 2.370 Keck PI program, and from the Carnegie Institution of Wash- 2452162.787 −8.300 2.370 ington. S.M. acknowledges support from the W. J. McDonald 2452445.988 17.880 2.100 Postdoctoral Fellowship. G.W.H. acknowledges support from 2452537.743 −4.900 2.340 NASA, NSF, Tennessee State University, and the State of Ten- 2452713.113 8.860 1.590 2452806.027 −4.580 1.990 nessee through its Centers of Excellence program. The work 2452850.901 6.090 2.040 herein is based on observations obtained at the W. M. Keck 2453179.985 12.360 1.730 Observatory, which is operated jointly by the University of 2453479.066 8.000 1.210 California and the California Institute of Technology, and we 2453549.857 −2.830 0.880 thank the UC-Keck, UH, and NASA-Keck Time Assignment 2453604.881 −2.530 1.220 Committees for their support. This research has made use of the 2453838.109 3.550 1.220 Keck Observatory Archive (KOA), which is operated by the W. 2453932.905 3.890 1.380 M. Keck Observatory and the NASA Exoplanet Science Insti- 2453960.873 3.880 1.630 tute (NExScI), under contract with the National Aeronautics and 2453961.820 −0.510 1.800 − Space Administration. We thank those who collected the data in 2453981.789 14.390 1.190 2453982.904 −1.880 1.900 the KOA including: D. Fischer, G. Marcy, J. Winn, W. Boruki, 2453983.831 −8.010 1.220 G. Bakos, A. Howard, and T. Bida. We also wish to extend our 2453984.887 −0.360 1.290 special thanks to those of Hawaiian ancestry on whose sacred 2454248.023 7.800 2.090 mountain of Mauna Kea we are privileged to be guests. With- 2454248.990 8.410 2.030 out their generous hospitality, the Keck observations presented 2454249.945 −3.960 1.730 herein would not have been possible. This research has made use 2454252.032 −12.260 1.870 of the SIMBAD database, operated at CDS, Strasbourg, France. 2454255.924 −12.790 1.280 This paper was produced using B AM . 2454277.851 8.200 1.500 Facilities: Keck:I (HIRES), APF (Levy Spectrometer), HET 2454278.896 11.300 1.700 (High-Resolution-Spectrograph) 2454279.933 11.850 1.580 2454294.903 −10.850 1.260 2454304.888 −4.110 1.570 APPENDIX 2454306.037 2.450 1.600 2454307.015 −2.050 1.500 Doppler Radial Velocity Observations7 2454308.072 5.540 1.600 2454309.050 7.370 1.810 Tables 4–6 present the collection of RV data used in the 2454310.042 8.730 1.720 detection of GL687. All data presented here have been corrected 2454311.024 1.600 1.480 to the solar system’s barycentric reference frame. 2454312.017 0.860 1.590 2454312.875 2.350 1.450 2454313.875 −1.170 1.320 2454314.908 6.220 1.630 7 Observations are corrected to the solar system’s barycentric reference 2454318.921 5.360 1.560 frame. 2454335.826 −13.560 1.940 2454338.888 −16.490 1.790 2454339.883 −9.590 1.780 Table 4 2454343.791 −8.170 1.950 HIRES/Keck Radial Velocities for GJ 687 2454396.717 3.080 2.390 JD RV Uncertainty 2454397.729 −4.380 2.130 (m s−1)(ms−1) 2454548.052 12.600 1.330 2454549.092 3.010 1.280 2450603.965 −15.930 1.770 2454633.963 −6.780 1.570 2450956.065 −4.740 2.040 2454634.903 −5.330 1.650 2450982.977 −12.880 1.560 2454635.946 −2.500 1.790 2451013.879 −15.550 1.610 2454636.901 2.170 1.650 2451312.036 −6.870 1.910 2454637.939 −5.830 1.610 2451368.798 −17.330 1.790 2454638.838 −5.010 1.450 2451439.745 −25.500 2.170 2454639.998 3.970 1.850 2451704.957 −14.920 1.810 2454641.954 2.760 1.540 12 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al.

Table 4 Table 6 (Continued) APF Radial Velocities for GJ 687

JD RV Uncertainty JD RV Uncertainty − − (m s−1)(ms−1) (m s 1)(ms1) 2454674.860 −0.820 1.440 2456484.738 11.650 0.910 2454688.892 −4.450 1.710 2456486.839 9.500 0.950 2454689.917 6.340 1.890 2456488.845 11.640 1.120 2454717.853 4.220 2.000 2456490.726 8.300 1.260 2454718.914 −2.870 2.070 2456492.816 7.020 0.900 2454719.866 1.210 2.040 2456494.776 5.260 0.750 2454720.871 0.570 2.140 2456508.764 −8.650 0.780 2454721.879 4.620 2.040 2456510.800 −9.020 0.800 2454722.797 5.810 2.210 2456513.772 −8.920 0.860 2454723.814 2.610 2.120 2456521.823 −2.120 0.730 2454724.848 1.350 1.990 2456522.790 −2.920 0.930 2454968.017 5.210 0.970 2456533.797 2.910 0.820 2455016.035 0.730 1.370 2456551.743 −8.410 0.810 2455022.084 1.680 1.670 2456552.717 −6.660 0.970 2455023.044 6.360 1.070 2456553.727 −7.610 0.850 2455024.823 −11.730 0.920 2456554.720 −4.640 0.890 2455049.825 0.220 1.090 2456564.762 1.530 0.940 2455050.858 3.990 1.650 2456568.692 0.460 1.140 2455051.772 6.360 1.790 2456571.657 0.580 0.940 2455052.780 8.540 1.930 2456622.032 −9.750 1.280 2455053.925 −12.540 1.510 2456677.921 2.390 0.980 2455143.692 −3.080 2.620 2456681.011 4.660 1.360 2455167.704 3.210 1.710 2455259.119 3.760 1.610 2455260.144 0.000 1.470 REFERENCES 2455371.991 −5.030 1.070 2455407.955 −1.480 1.060 Anglada-Escude,´ G., Arriagada, P., Vogt, S. S., et al. 2012, ApJL, 751, L16 2455463.718 −8.590 1.520 Balan, S. T., & Lahav, O. 2009, MNRAS, 394, 1936 2455516.696 −1.120 1.740 Baluev, R. V. 2012, MNRAS, 422, 2372 2455518.722 −3.780 2.320 Baraffe, I., Chabrier, G., Allard, F., & Hauschildt, P. H. 1998, A&A, 337, 403 Batalha, N. M., Rowe, J. F., Bryson, S. T., et al. 2013, ApJS, 204, 24 2455609.135 11.250 1.310 − Boisse, I., Bonfils, X., & Santos, N. C. 2012, A&A, 545, A109 2455638.121 3.930 0.910 Boyajian, T. S., von Braun, K., van Belle, G., et al. 2012, ApJ, 757, 112 − 2455639.143 2.250 0.910 Butler, R. P., Marcy, G. W., Williams, E., et al. 1996, PASP, 108, 500 2455665.046 −5.130 1.190 Butler, R. P., Vogt, S. S., Marcy, G. W., et al. 2004, ApJ, 617, 580 2455670.082 −8.090 1.290 Charbonneau, P. 2010, LRSP, 7, 3 2455721.057 6.840 1.850 Cutri, R. M., Skrutskie, M. F., van Dyk, S., et al. 2003, 2MASS All Sky Catalog 2455825.832 3.030 2.280 of Point Sources (Pasadena, CA: NASA/IPAC Infrared Science Archive) 2455840.748 3.820 1.390 Dressing, C. D., & Charbonneau, D. 2013, ApJ, 767, 95 2456027.098 11.390 1.940 Dumusque, X., Pepe, F., Lovis, C., et al. 2012, Natur, 491, 207 Eaton, J. A., Henry, G. W.,& Fekel, F. C. 2003, in The Future of Small Telescopes 2456116.938 16.010 1.130 In The New Millennium, ed. T. D. Oswalt (Astrophysics and Space Science 2456117.906 14.270 1.050 Library, Vol. 288; Dordrecht: Kluwer), 189 2456329.167 5.420 1.970 Efron, B. 1979, AnSta, 7, 1 2456432.892 −0.700 1.300 Endl, M., Cochran, W. D., Tull, R. G., & MacQueen, P. J. 2003, AJ, 126, 3099 2456433.935 1.610 1.340 Fischer, D. A., & Valenti, J. 2005, ApJ, 622, 1102 2456548.855 −2.340 1.490 Ford, E. B. 2005, AJ, 129, 1706 2456549.793 −7.780 1.430 Ford, E. B., & Gaudi, B. S. 2006, ApJL, 652, L137 2456550.849 −3.840 2.150 Gonzalez, G. 1997, MNRAS, 285, 403 2456551.724 −2.520 1.570 Gregory, P. C. 2011, MNRAS, 415, 2523 Henry, G. W. 1999, PASP, 111, 845 Henry, G. W., Fekel, F. C., & Hall, D. S. 1995, AJ, 110, 2926 Henry, G. W., Fekel, F. C., Kaye, A. B., & Kaul, A. 2001, AJ, 122, 3383 Henry, T. J., & McCarthy Jr., D. W. 1993, AJ, 106, 773 Jenkins, J. S., Jones, H. R. A., Tuomi, M., et al. 2013, ApJ, 766, 67 Table 5 Jenkins, J. S., Ramsey, L. W., Jones, H. R. A., et al. 2009, ApJ, 704, 975 Hobby–Eberly Telescope Radial Velocities for GJ 687 Johnson, J. A., & Apps, K. 2009, ApJ, 699, 933 Kukla, A. 2010, Extraterrestrials: A Philosophical Perspective (Lanham, MD: JD RV Uncertainty Lexington Books) − − (m s 1)(ms1) Lissauer, J. J., Ragozzine, D., Fabrycky, D. C., et al. 2011, ApJS, 197, 8 2452394.962 −2.010 2.220 Mayor, M., Marmier, M., Lovis, C., et al. 2011, arXiv:1109.2497 Mayor, M., Udry, S., Lovis, C., et al. 2009, A&A, 493, 639 2452395.925 −3.600 2.600 Meschiari, S., Wolf, A. S., Rivera, E., et al. 2009, PASP, 121, 1016 − 2452396.936 1.400 2.870 Meschiari, S., Wolf, A. S., Rivera, E., et al. 2012, ascl soft, 1210.018 2452400.920 3.800 2.880 Neves, V., Bonfils, X., Santos, N. C., et al. 2013, A&A, 551, A36 2452403.869 1.370 2.290 Paulson, D. B., Saar, S. H., Cochran, W. D., & Henry, G. W. 2004, AJ, 127, 1644

13 The Astrophysical Journal, 789:114 (14pp), 2014 July 10 Burt et al.

Petigura, E. A., Howard, A. W., & Marcy, G. W. 2013, PNAS, 110, 19273 Sousa, S. G., Santos, N. C., Israelian, G., Mayor, M., & Udry, S. 2011, A&A, Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. 1992, 533, A141 Numerical Recipes in C: The Art of Scientific Computing (2nd ed.; Triaud, A. H. M. J., Anderson, D. R., Collier Cameron, A., et al. 2013, A&A, Cambridge: Cambridge Univ. Press) 551, A80 Queloz, D., Henry, G. W., Sivan, J. P., et al. 2001, A&A, 379, 279 Vogt, S. S., Allen, S. L., Bigelow, B. C., et al. 1994, Proc. SPIE, 2198, 362 Rivera, E. J., Laughlin, G., Butler, R. P., et al. 2010, ApJ, 719, 890 Vogt, S. S., Butler, R. P., Rivera, E. J., et al. 2014a, ApJ, 787, 97 Roberts, D. H., Lehar, J., & Dreher, J. W. 1987, AJ, 93, 968 Vogt, S. S., Radovan, M., Kibrick, R., et al. 2014b, ApJ, 787, 97 Rojas-Ayala, B., Covey, K. R., Muirhead, P. S., & Lloyd, J. P. 2012, ApJ, Wright, J. T. 2005, PASP, 117, 657 748, 93 Wright, J. T., Fakhouri, O., Marcy, G. W., et al. 2011, PASP, 123, 412 Scargle, J. D. 1982, ApJ, 263, 835 Wright, J. T., Marcy, G. W., Butler, R. P., & Vogt, S. S. 2004, ApJS, Schlaufman, K. C., & Laughlin, G. 2010, A&A, 519, A105 152, 261

14